A deep learning framework for optimizing personalized online course recommendation and selection

Massive Open Online Courses (MOOCs) and the broad adoption of distance learning over the past few years have caused a remarkable shift in the educational landscape. However, the vast majority of available MOOCs often challenge learners in selecting courses that align with their academic goals, leading to high dropout rates. This paper presents an unexploited opportunity for formal educational institutions to integrate MOOCs into their curricula by guiding and supporting learners on these platforms. It introduces a Deep Semantic MOOC Recommender System (DSMRS) designed to help learners choose MOOCs aligned with their formal curriculum. The system utilizes advanced Natural Language Processing (NLP) techniques to deliver personalized recommendations to learners. It employs a top-N recommender algorithm and leverages three key strategies: (a) an optimized Explicit Semantic Analysis (ESA) to measure semantic similarity between course descriptions and learning objectives; (b) Sentiment Analysis, using a Bayesian Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model to analyze learner reviews of MOOCs; and (c) Classification of Recommended MOOCs based on Bloom's Taxonomy, categorizing MOOCs according to cognitive complexity. The results highlight that experimentation conducted with the system demonstrates promising performance.